View Full Version : Accuracy vs Std Deviation: Conflicting Data

For the past few months I had been using 4.0 gr of VV N320 behind a 124gr FMJ Zero bullet as my USPSA Production load. I decided to give this load a try 'cause a lot of people told me that VV was the cat's meow and that if I was using it, I was not one of the cool kids.

What I have found out is that, when I have chrono'd N320, it has a very small Extreme Spread and Standard Deviation but today I compared my results against Titegroup and I was surprised by what I discovered. It turns out that, even though the ES & SD of Titegroup is not as tight as that of N320, the accuracy of TG was better.

I was under the impression that a tight ES/SD was a good indicator of the accuracy of that load but it seems that this is not always the case. Is this a common issue or am I interpreting my results all wrong?

Yep your not the first person to observe this. Reloading to achieve low ES and SD is ok if that's your goal but i always reload for best accuracy. And they rarely come in the same load. But who cares? You will never see the effects of a large ES or SD at pistol range distance. Load for best accuracy. Now rifle reloading, that's where ES and SD start to matter.

Outdoor Hub mobile, the outdoor information engine

You will never see the effects of a large ES or SD at pistol range distance.

Now rifle reloading, that's where ES and SD start to matter.

That makes sense. Thank you!

nope for the most part is doesn't matter. now if you ES is really out of wack, like 150 or 200 fps the longer the range the larger the groups. but if you're ES is that large something isn't right anyway.

nope for the most part is doesn't matter. now if you ES is really out of wack, like 150 or 200 fps the longer the range the larger the groups. but if you're ES is that large something isn't right anyway.

I don't remember what the ES was between N320 and TG but I do remember that the SD of N320 was in the neighborhood of 6 and TG was closer to 15 which seems like a huge difference.

fredj338

11-27-2011, 10:34

I don't remember what the ES was between N320 and TG but I do remember that the SD of N320 was in the neighborhood of 6 and TG was closer to 15 which seems like a huge difference.

Actually, SD in the teens is very good. As noted though, doesn't mean a lot as to close range accuracy (100yds or less). At ranges beyond 200yds, wide vel spreads wil obviously affect accuracy as the bullets are arriving on target & vastly diff vel, dorp & wind will play a bigger role. Some of my best shooting handgun ammo will have SD as high as 30, still shoots into one hole @ 50ft. The vast majority of shooters can't taake advantage of super accurate ammo anway. Unless you & your gun are already capable of sub 1 1/2" groups @ 75ft, you would never be able to establish match grade accurate ammo from accurate ammo.:dunno:

With VV N320, my SDs were 5 or so for my 9mm loads with Precision Delta bullets. With TG, still well below 10, more like 7-8.

Three-Five-Seven

11-27-2011, 13:02

I've been working up a new 1000 yard rifle in .260 Remington.

Had a good trend going, with decreasing extreme spreads as the load intensity increased. Got the E.S. down in the low to mid teens. Two loads shot great, but the one in between those two was terrible.

Guys at the range want me to believe that barrel vibration at different intensities is variable enough to cause these results.

So, apparently, low extreme spreads -- in, and of themselves -- is not the whole story for accuracy.

Guns talk, it's our job to listen to them.

fredj338

11-27-2011, 15:32

I've been working up a new 1000 yard rifle in .260 Remington.

Had a good trend going, with decreasing extreme spreads as the load intensity increased. Got the E.S. down in the low to mid teens. Two loads shot great, but the one in between those two was terrible.

Guys at the range want me to believe that barrel vibration at different intensities is variable enough to cause these results.

So, apparently, low extreme spreads -- in, and of themselves -- is not the whole story for accuracy.

Guns talk, it's our job to listen to them.

When you say "intensity" do you mean higher vel levels? I find most powders just get better as the load density goes up, particularly slower powders. Yes, bbl vibration, particularly longer bbls, can play havoc w/ accuracy. All rifles have sweet spots w/ their given bullet/powder/velocity combo. I am shooting a 27" 260AI, terrific caliber, very accurate, no recoil, flatter shooting than a 308, but alas, I have only been able to run it out to 300yds at my range.:crying: I have had great 300yd accuracy w/ some loads that are in single dig SD & some in the teens. Not many will shoot well once the SD gets into the high 20s & up, especially on windy days.

Three-Five-Seven

11-27-2011, 15:45

Yes, we are VERY fortunate to have access to a 1000 yard range!

Well, I was (trying) using the word "intensity" to mean: more powder = more velocity = more pressure.

After a couple of years of this long range stuff with smokeless powder rounds, I've learned that the highest achievable velocity is not the best for cases, the gun, or me. So, lower "intensity" rounds often perform as well, or better than the hottest loads that can be done and still have the rifle in one piece at the end of the day. The intensity is the commotion I feel when touching hot rounds off -- to say nothing of the attendant barrel wear, heat, noise, and shortened component life.

The powder I was referring to above is H4831sc, which is a relatively slow burning powder. My starting loads had a very high extreme spread (40 fps), which quickly dropped as "intensity" (amount of powder) increased.

What bullets have worked best in your 6.5??? I've tried Amax so far because I've had such good luck with them in the 308. Gonna try some Hornady BTHP 140 grain bullets next. What's worked well in your Ackley?

All the best.

fredj338

11-27-2011, 17:44

Yes, we are VERY fortunate to have access to a 1000 yard range!

Well, I was (trying) using the word "intensity" to mean: more powder = more velocity = more pressure.

After a couple of years of this long range stuff with smokeless powder rounds, I've learned that the highest achievable velocity is not the best for cases, the gun, or me. So, lower "intensity" rounds often perform as well, or better than the hottest loads that can be done and still have the rifle in one piece at the end of the day. The intensity is the commotion I feel when touching hot rounds off -- to say nothing of the attendant barrel wear, heat, noise, and shortened component life.

The powder I was referring to above is H4831sc, which is a relatively slow burning powder. My starting loads had a very high extreme spread (40 fps), which quickly dropped as "intensity" (amount of powder) increased.

What bullets have worked best in your 6.5??? I've tried Amax so far because I've had such good luck with them in the 308. Gonna try some Hornady BTHP 140 grain bullets next. What's worked well in your Ackley?

All the best.

H4831sc is a good perfromer in my 260AI w/ medium to hvy, 120gr-142gr. I have a 1-8 twist, so it favors the 139gr-142gfr bullets, but is no slouch w/ even the 85gr Sierra. I can get into the 0.3s w/ them as well as the 142grSMK. The A-max is also a good choice in 120gr or 140gr. I also like the 139gr Lapua. My fav powder for max vel w/ exc accuracy is 7828ssc, but H4831sc & RL22 also do a great job. I can get 2800fps & not push max. The AI gives about 1.5gr more powder volume & almost no case stretch. Also give H4350 a try. It will drop about 75fps but tremendous accuracy w/ 85-120gr bullets & does very we;; w/ the 140gr but @ 2700fps+.

This is not suprising.

You must understand that when you are chronographing loads, you are not just measuring the bullet. You are measuring the bullet speed and the accuracy of measuring equipment.

The first question that I would ask, is how do you know that you are using the right distribution? When people talk about "standard deviation" that assumes a normal distribution. Do you have enough data to show it is normal? Maybe you are loading in a region where it follows a weibull distribution and you are comapring sigma (standard deviation of a normal distribution) with the k and lamda of a weibull?

The short version of this is that before numbers are used, understand them. 20 years ago, very few people would have talked about standard deviation because they could not even compute it. Now, a simple program computes it. It doesn't mean its a valid value to use.

So shoot what shoots best on the target and let the rest of the calculated numbers fall where they fall (unless its pressure and then pay attention).

-Dana

HAMMERHEAD

11-30-2011, 16:09

In 9mm SR 7625 is another powder that doesn't have the smallest SD numbers, but it almost always is the most accurate in my 9's.

Sd's usually run around 50-60 fps with 7625.

In 9mm SR 7625 is another powder that doesn't have the smallest SD numbers, but it almost always is the most accurate in my 9's.

Sd's usually run around 50-60 fps with 7625.

50-60 fps is ES, Extreme Spread, not standard deviation. SD is not expressed in fps.

Your SD number will most likely be 15 or less I would think.

Three-Five-Seven

11-30-2011, 18:08

Good gravy, there are so many horse feathers flying around in that you can't see across the room.

Standard deviation is the sum of the squared deviations from the mean, divided by the number of observations. It has nothing to do with the distribution of data. Other statistical metrics describe that. And, S.D. is definitely expressed in what ever unit is being observed, be it Kumquats or Xylophones.

Good gravy, there are so many horse feathers flying around in that you can't see across the room.

Standard deviation is the sum of the squared deviations from the mean, divided by the number of observations. It has nothing to do with the distribution of data. Other statistical metrics describe that. And, S.D. is definitely expressed in what ever unit is being observed, be it Kumquats or Xylophones.

See, this is why I say it dangerous to give numbers that are easy to calculate.

My stats books has the following:

"The population standard deviation is an important index of variability that conveys a great deal of information. It is a fundamental parameter of the mathematical function the normal curve that fits the frequency distribution of so many populations."

Sorry, but standard deviation is based upon the normal distribution. All other distributions you can "estimate" a lower bounds on the probability of a random variable falls with +/-k standard deviations using Chebyshev's Theorem. But it is always better to use the correct distribution.

The next part is that you are finding the sample standard deviation. You have translate that into a population standard deviation.

Now separate out the various components of the variability. Typically this is done using a repeatability and reproducibility study. Again, it can be done, but how many can tell us the R&R of their chronograph? It has an inherent measurement variability that is hidden in the sample standard deviations

.

Me. The chrono data is fun for energy and making sure loads aren't too hot. If the load is within the velocity envelope that you want (hopefully safe limits) and it shoots well, us it.

-Dana

fredj338

11-30-2011, 20:56

Good gravy, there are so many horse feathers flying around in that you can't see across the room.

Standard deviation is the sum of the squared deviations from the mean, divided by the number of observations. It has nothing to do with the distribution of data. Other statistical metrics describe that. And, S.D. is definitely expressed in what ever unit is being observed, be it Kumquats or Xylophones.

I'm glad you posted that, I thought I was the only one confused.:yawn:

GioaJack

11-30-2011, 20:57

Well, that clears up any confusion on the matter. :faint:

Jack

Well, that clears up any confusion on the matter. :faint:

Jack

Well. I would talk to my wife tonight instead of myself but she is being a PITA. She is complaining (the GT word not the one i would use) about me not having a job. So, a place wants to once again pay me obscenely to do some part time work. Problem is, I need to discuss the deal over some beers so they invited my to do that. She is agitated about me having to go there and not to her graduation for her masters.

So, that means you guys get me tonight.

Nothing I love more than equations (well other than: shooting, beer, fast cars, women, motorcycles).

Last weekend I got into modeling the spring rate to the cycling rate of glocks. My equations worked.

Time to go load. Of do some statistics....or something.

-Dana

Well, that clears up any confusion on the matter. :faint:

Jack

But you have a bigger trailer. Do you have a relaoding room in it (there is a good idea!!)

-Dana

Spending just a few minutes reviewing old articles about using a chronograph, it is quickly clear that a low SD has almost no effect on accuracy (doesn't matter how much you think it should, it just doesn't).

It may play a part for 600+ yd shooting, but certainly NOT for handguns at ranges out to 50 yards.

About the only thing a chronograph is really good for is meeting a power factor. For rifles, you can calculate external ballistic data and determine the standard error of the trajectory calculations.

Spending just a few minutes reviewing old articles about using a chronograph, it is quickly clear that a low SD has almost no effect on accuracy (doesn't matter how much you think it should, it just doesn't).

It may play a part for 600+ yd shooting, but certainly NOT for handguns at ranges out to 50 yards.

About the only thing a chronograph is really good for is meeting a power factor. For rifles, you can calculate external ballistic data and determine the standard error of the trajectory calculations.

Since I barely got through statistics in College, I will not go there-

As always though, when reviewing the accuracy of handgun loads, you must eliminate the human factor from the shooting. You are going to need a Ransom rest or equivalent.

Since I barely got through statistics in College, I will not go there-

As always though, when reviewing the accuracy of handgun loads, you must eliminate the human factor from the shooting. You are going to need a Ransom rest or equivalent.

Yes/No.

To determine actual mechanical accuracy yes. But keep in mind that different powders will have different recoil impulses. This can actually affect the shooter and the shooter may shoot one than another even though pure mechanical accuracy (i.e ransom rest) may say otherwise.

I suspect that at 100 meters, that most people will shoot a 17HMR that has a mechanical accuracy of 0.5MOA better than 338 Rem UM with 0.5MOA.

-Dana

HAMMERHEAD

12-01-2011, 16:40

50-60 fps is ES, Extreme Spread, not standard deviation. SD is not expressed in fps.

Your SD number will most likely be 15 or less I would think.

You're right, my bad.

Yes/No.

To determine actual mechanical accuracy yes. But keep in mind that different powders will have different recoil impulses. This can actually affect the shooter and the shooter may shoot one than another even though pure mechanical accuracy (i.e ransom rest) may say otherwise.

I suspect that at 100 meters, that most people will shoot a 17HMR that has a mechanical accuracy of 0.5MOA better than 338 Rem UM with 0.5MOA.

-Dana

I see your point Dana. There can also be psychological/physical factors that determine accuracy. For example, did you begin to flinch after firing the first rounds? Do you in your mind just believe one is better then the other and have this affect the shot string? Did the weather, amount of sun, light in the room change? Did the person next to you in the firing line begin to pop off rounds during your shot string? Most of these factors could be brought under control with the mechanical rest.

You're right, my bad.

I was correct only in saying that I suspected you were talking about ES and not SD with those numbers. But I was wrong in saying SD was not expressed in the same units as what is being measured. It is fps, at least in this case.

vBulletin® v3.8.7, Copyright ©2000-2014, vBulletin Solutions, Inc.